As AI systems are playing a larger and larger role in decision-making, in contexts such as hiring decisions, loan applications, and legal verdicts, it’s increasingly critical to ensure that they are fair. In this paper, causaLens Principal Investigator Hana Chockler together with Professor Joe Halpern of Cornell University, develop novel causal methods for evaluating fairness. This includes an analysis of the computational complexity of deciding fairness, and a practical regulatory framework for determining disparate impact. The paper has been accepted to the 36th AAAI Conference on Artificial Intelligence (AAAI’22), a top conference in AI.
